Practice English vocabulary for maintaining knowledge graphs: entity updates, refresh schedules, orphaned node cleanup, entity resolution, and graph health scores.
0 / 5 completed
1 / 5
What does 'the entity is updated when the source data changes' mean?
Knowledge graph freshness depends on change detection and propagation. This can be triggered by CDC (Change Data Capture) streams, periodic batch jobs comparing source hashes, or webhooks from source systems notifying the graph of updates.
2 / 5
What is 'the graph refresh schedule' for a knowledge graph?
Refresh schedules balance data freshness needs against the cost of re-ingestion. Critical entities (product prices, news articles) may refresh hourly; stable entities (historical facts, geographic data) may only need weekly updates.
3 / 5
What are 'orphaned nodes' and why are they cleaned up?
Orphaned nodes accumulate when source records are deleted but the graph cleanup job doesn't run, or when relationship data changes. They waste storage, distort graph metrics, and can surface in queries as seemingly valid entities with no useful connections.
4 / 5
What is 'entity resolution' that 'merges duplicates'?
Entity resolution (also called entity matching or deduplication) uses similarity algorithms, unique identifiers, and ML models to recognize that 'Apple Inc.', 'Apple Computer', and 'AAPL' all refer to the same entity — then merging them into one node with all related data.
5 / 5
What does 'the graph health score tracks data freshness and completeness' measure?
Graph health scores give operators a single number representing graph quality. Freshness measures how recently data was updated (e.g., % of entities updated in the last 7 days). Completeness measures how many required fields are populated. Low scores trigger investigation and remediation.